Space-based full-link photoelectric imaging simulation signal generation method based on CUDA (Compute Unified Device Architecture)

Abstract

The invention provides a space-based full-link photoelectronic imaging simulation method based on CUDA (Compute Unified Device Architecture), and solves the problems of complicated calculation, insufficient simulation precision and high hardware requirements in the prior art. The method comprises the following steps: dividing the surface of a target three-dimensional model into triangular surface elements and calculating target intrinsic radiation scattering data of each surface element; generating background intrinsic radiation scattering data in real time through CUDA (Compute Unified Device Architecture) parallel calculation according to the space-time spectrum parameters and the earth surface / cloud layer / limb basic model library; calculating characterization radiation before entrance pupil based on target and background intrinsic data; the entrance pupil radiation is converted into a detector voltage signal through CUDA parallel optimization, and a gray infrared simulation image is generated. According to the method, the calculation efficiency is remarkably improved, and space-based full-link photoelectric imaging simulation with high authenticity and real-time performance is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of space exploration, and in particular to a space-based full-link photoelectric imaging simulation signal generation method based on CUDA. BACKGROUND

[0002] Compared with traditional land, sea and air-based detection systems, space-based detection systems have the advantages of large observation field of view, wide detection range, no influence of bad weather, all-weather operation, high target detection sensitivity, etc. In the field of space-based infrared detection, the United States has always been at the forefront of world development. In recent years, scientific research has increasingly demanded infrared detection of different weather, different scenes and various types of targets, but field simulation research is very dependent on meteorological conditions and instrument conditions, requiring a large amount of manpower, material resources and time cost. Therefore, the field of space-based infrared detection needs to establish an imaging simulation model to accurately detect and calculate images of targets and backgrounds in different scenes. It is of great significance to establish a full-link imaging model with high simulation accuracy and high real-time performance for space-based infrared field detection simulation imaging.

[0003] In terms of imaging simulation models, Xu Yuxin et al. proposed a simulation analysis technology for space target infrared properties, considering complex environmental radiation sources and the radiation characteristics of the target itself and various actual influencing factors in the imaging process, such as satellite orbit parameters, detector noise effects, etc., improving the realism of the simulation scene. Yuan Hang et al. established an optical radiation characteristic model of the complex environmental element comprehensive action of the aircraft exhaust plume, forming a full-digital link spectral radiation imaging characteristic accurate prediction model coupling elements such as aircraft exhaust plume, sea surface / cloud background, environmental atmosphere, optical system and imaging detector.

[0004] In terms of imaging efficiency of space-based detection systems, Xu Feifan used GPU (Graphics Processing Unit) parallel computing acceleration to develop an infrared program simulation software for parallel computing through the OptiX ray tracing engine, and simulated and calculated different scenes in the mid-infrared and far-infrared bands, which can quickly generate accurate simulation images.

[0005] In the prior art, the research on space-based full-link photoelectric imaging simulation systems is relatively rich in the study of infrared characteristics of targets and backgrounds in various bands, but most of them are based on infrared radiation transmission theory to locally optimize infrared image calculation, rarely discuss the establishment of imaging detector models and consider various physical effects of sensors in the full-link simulation process, and most of the research does not make good use of GPU parallel computing to improve computing efficiency.

[0006] The existing space-based full-link simulation imaging technology is mostly based on an infrared radiation transmission model, and a space-based detection simulation full-link imaging model is established by calculating the infrared imaging characteristics of a target and infrared radiation data of a detection background environment. In actual application, the space-based full-link imaging model covers optical signal generation and propagation, photoelectric conversion, and electrical signal processing, and the actual calculation process is complex, the simulation accuracy and realism still need to be improved, and a large amount of image matrix calculation is accompanied in the process, which requires a high-performance computer device, and the actual imaging real-time performance is not ideal only by using a CPU. SUMMARY

[0007] The application provides a CUDA-based space-based full-link photoelectric imaging simulation signal generation method, which solves the problems of complex actual calculation process, low simulation accuracy and realism, and high requirement for a computer device in the prior art, effectively improves the calculation efficiency of space-based full-link photoelectric imaging simulation, and realizes high realism and real-time performance of a space-based full-link photoelectric imaging simulation image.

[0008] The application provides a CUDA-based space-based full-link photoelectric imaging simulation signal generation method, which solves the problems of complex actual calculation process, low simulation accuracy and realism, and high requirement for a computer device in the prior art, effectively improves the calculation efficiency of space-based full-link photoelectric imaging simulation, and realizes high realism and real-time performance of a space-based full-link photoelectric imaging simulation image. The surface of a target three-dimensional model is divided into a plurality of triangular surface elements, target intrinsic radiation scattering data of each triangular surface element is calculated, and background intrinsic radiation scattering data is generated in real time through CUDA parallel calculation according to space-time spectrum parameters, ground / cloud / limb basic data and a model library; wherein the target intrinsic radiation scattering data includes intrinsic radiance and target total radiation intensity of each surface element. Target pre-pupil characteristic radiation and background pre-pupil characteristic radiation are calculated according to the target intrinsic radiation scattering data and the generated background intrinsic radiation scattering data. The target pre-pupil characteristic radiation and the background pre-pupil characteristic radiation are converted into detector voltage signals by using CUDA parallel calculation optimization, and a gray infrared simulation image is obtained.

[0009] In a possible implementation manner, the surface of the target three-dimensional model is divided into a plurality of triangular surface elements, and the target intrinsic radiation scattering data of each triangular surface element is calculated, including: Determination of observation parameters of each triangular surface element; wherein the observation parameters include surface element area, surface element normal and surface element temperature; According to the surface element temperature, the intrinsic radiance of each triangular surface element is obtained by using a radiation brightness calculation formula; Under different wave bands, a mapping table under different observation angles and solar azimuth angles is obtained based on a solar radiation model; wherein the mapping table is a mapping of target-to-camera atmospheric transmittance and solar reflection value. The CUDA parallel calculates a first observation angle of each triangular facet according to an included angle between the facet normal and a line-of-sight direction of a camera observation facet element; The CUDA parallel calculates a first zenith angle of each triangular facet according to an included angle between the facet normal and a direction in which a sunlight ray direction points to the facet element; The CUDA parallel calculates a first observation angle of each triangular facet according to an included angle between the facet normal and a line-of-sight direction of a camera observation facet element; The CUDA parallel calculates a first observation angle of each triangular facet according to an included angle between the facet normal and a line-of-sight direction of a camera observation facet element;

[0010] In a possible implementation, the CUDA parallel calculation is used to generate background intrinsic radiance scattering data in real time according to spatiotemporal spectral parameters, ground surface / cloud layer / edge basis data and a model library, and the method comprises the following steps: The CUDA parallel calculates a first observation angle of each triangular facet according to an included angle between the facet normal and a line-of-sight direction of a camera observation facet element; The CUDA parallel calculates a first observation angle of each triangular facet according to an included angle between the facet normal and a line-of-sight direction of a camera observation facet element; The CUDA parallel calculates a first observation angle of each triangular facet according to an included angle between the facet normal and a line-of-sight direction of a camera observation facet element; The CUDA parallel calculates a first observation angle of each triangular facet according to an included angle between the facet normal and a line-of-sight direction of a camera observation facet element; The CUDA parallel calculates a first observation angle of each triangular facet according to an included angle between the facet normal and a line-of-sight direction of a camera observation facet element; The CUDA parallel calculates a first observation angle of each triangular facet according to an included angle between the facet normal and a line-of-sight direction of a camera observation facet element; The CUDA parallel calculates a first observation angle of each triangular facet according to an included angle between the facet normal and a line-of-sight direction of a camera observation facet element; The CUDA parallel calculates a first observation angle of each triangular facet according to an included angle between the facet normal and a line-of-sight direction of a camera observation facet element; The CUDA parallel calculates a first observation angle of each triangular facet according to an included angle between the facet normal and a line-of-sight direction of a camera observation facet element; The CUDA parallel calculates a first observation angle of each triangular facet according to an included angle between the facet normal and a line-of-sight direction of a camera observation facet element; The CUDA parallel calculates a first observation angle of each triangular facet according to an included angle between the facet normal and a line-of-sight direction of a camera observation facet element.

[0011] In a possible implementation, the target pre-pupil characteristic radiation is represented as: ; The background pre-entry characteristic radiation is represented as: ; Wherein, represents target intrinsic radiation scattering; represents atmospheric transmittance from the target to the detector; represents ground surface pre-entry characteristic radiation; represents cloud layer pre-entry characteristic radiation.

[0012] In a possible implementation, the conversion of the target pre-entry characteristic radiation and the background pre-entry characteristic radiation into a detector voltage signal by using CUDA parallel computing optimization to obtain a grayscale infrared simulation image comprises: superimposing and fusing the target pre-entry characteristic radiation and the background pre-entry characteristic radiation to obtain input radiation; constructing an imaging detector model to perform photoelectric detector signal level imaging modeling on the input radiation to obtain a grayscale infrared simulation image.

[0013] In a possible implementation, the imaging detector model comprises an optical system module, a detector module, and a signal processing module; The optical system module is configured to perform energy transmission and spatial characteristic conversion operations on the input radiation by optical transmittance, spatial modulation, and distortion correction mechanisms to obtain image plane modulation radiation brightness; The detector module is configured to perform optical band conversion, space-time domain filtering, and noise superposition on the image plane modulation radiation brightness to obtain a noisy detector voltage signal; The signal processing module is configured to perform circuit noise superposition and grayscale quantization on the noisy detector voltage signal to obtain a grayscale infrared simulation image.

[0014] In a possible implementation, the conversion of the target pre-entry characteristic radiation and the background pre-entry characteristic radiation into a detector voltage signal by using CUDA parallel computing optimization to obtain a grayscale infrared simulation image comprises: performing line-of-sight direction offset calculation by adding distortion parameters to obtain corrected second observation angles and geographic coordinates; re-performing the target pre-entry characteristic radiation and the background pre-entry characteristic radiation according to the corrected second observation angles to obtain updated target pre-entry characteristic radiation and background pre-entry characteristic radiation, and calculating updated input radiation according to the updated target pre-entry characteristic radiation and the background pre-entry characteristic radiation; performing band attenuation calculation on the updated input radiation according to an optical system transmittance database to obtain attenuated radiation; The MTF spatial modulation is performed on the attenuated radiation to obtain an image plane modulation radiation luminance.

[0015] In a possible implementation, the optical band conversion, space-time domain filtering and noise superposition on the image plane modulation radiation luminance to obtain a noisy detector voltage signal comprises: According to the detector physical parameters, photoelectric conversion is performed on the image plane modulation radiation luminance to obtain an initial voltage signal; According to the detector pixel size, a spatial transfer function is calculated to obtain a spatial transfer function, and according to the detector carrier lifetime, a time filtering function is calculated to obtain a time filtering function; The initial voltage signal is spatially filtered according to the spatial transfer function and the time filtering function to obtain a noisy detector voltage signal.

[0016] In a possible implementation, the circuit noise superposition and gray scale quantization on the noisy detector voltage signal to obtain a gray scale infrared simulation image comprises: According to the circuit design parameters, a Gaussian distribution is used to generate circuit noise; The circuit noise is superimposed on the noisy detector voltage signal to obtain a degree infrared simulation image.

[0017] One or more technical solutions provided in the present application have at least the following technical effects or advantages: The present application realizes target geometry high-precision modeling and radiation characteristic decoupling through triangular facet division, dynamically generates a realistic background radiation field in combination with space-time parameters and a basic model library; the CUDA parallel calculation is used to significantly improve the background radiation field generation and sensor signal conversion efficiency, and ensure the real-time performance of complex scene simulation; the target radiance and total radiation intensity are completely calculated to support multi-dimensional applications, the state of the target and background radiation after atmospheric transmission to the sensor entrance pupil is accurately simulated, and the optical effect and detector physical property simulation are efficiently completed, and finally a high-fidelity gray scale infrared image is output in real time. BRIEF DESCRIPTION OF DRAWINGS

[0018] Figure 1 A CUDA-based space-based full-link photoelectric imaging simulation signal generation method step flowchart is provided for the embodiments of the present application. Figure 2 A target / background intrinsic radiation generation module framework diagram based on CUDA is provided for the embodiments of the present application. Figure 3 A target / background entrance pupil pre-characteristic radiation generation module framework diagram based on CUDA is provided for the embodiments of the present application. Figure 4A detector response model and image generation module framework based on CUDA parallel computing are provided for an embodiment of the present application. Figure 5 A space-based infrared detection simulation platform framework based on CPU-GPU cooperative work of CUDA is provided for an embodiment of the present application. DETAILED DESCRIPTION

[0019] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0020] A space-based full-link photoelectric imaging simulation signal generation method based on CUDA, referring to Figure 1 , the method comprises the following steps S101 to S103.

[0021] S101, the target three-dimensional model surface is divided into a plurality of triangular surface elements, the target intrinsic radiation scattering data of each triangular surface element is calculated, and the background intrinsic radiation scattering data is generated in real time through CUDA parallel calculation according to the space-time spectrum parameters, the ground / cloud / sky boundary basic data and the model library; wherein the target intrinsic radiation scattering data comprises the intrinsic radiance of each surface element and the target total radiation intensity. In step S101, the target three-dimensional model surface is divided into a plurality of triangular surface elements, the target intrinsic radiation scattering data of each triangular surface element is calculated, referring to Figure 2 , comprising the following steps S1011 to S1017.

[0022] S1011, the observation parameters of each triangular surface element are determined; wherein the observation parameters comprise the surface element area, the surface element normal and the surface element temperature. S1012, the intrinsic radiance of each triangular surface element is obtained according to the surface element temperature by using the radiation brightness calculation formula. S1013, under different wave bands, the mapping table under different observation angles and solar azimuth angles is obtained based on the solar radiation model; wherein the mapping table is the mapping of the target-to-camera atmospheric transmittance and the solar reflection value. S1014, the CUDA parallel takes the included angle between the surface element normal and the line-of-sight direction of the camera observed surface element as the first observation angle of each triangular surface element. S1015, the CUDA parallel takes the included angle between the surface element normal and the direction of the sunlight line direction pointing to the surface element as the first zenith angle of each triangular surface element. S1016, according to the first observation angle and the first zenith angle in the mapping table, the solar reflection value of each triangular facet is obtained; S1017, the target intrinsic radiation scattering data of each triangular facet is calculated according to the solar reflection value.

[0023] Exemplarily, the target intrinsic radiation scattering generation process is roughly as follows: (1) Target micro-facet division: input the target three-dimensional geometric model and flow field, temperature field data, and divide the target three-dimensional geometric model into grids through the CFD (Computational Fluid Dynamics) software, and divide the target surface into a plurality of fine triangular facets, and the physical information contained in each fine triangular facet includes the facet area of the current facet, the facet normal of the current facet and the facet temperature of the current facet.

[0024] (2) CUDA parallel calculation of target intrinsic radiance: set the emissivity of each facet to be the same value, utilize the facet temperature of each facet, combine the current waveband range, and utilize the radiance calculation formula (Planck formula) to calculate the intrinsic radiation brightness of each facet .

[0025] (3) CUDA parallel calculation of target total radiation brightness, based on the solar reflection radiation model, utilize the Modtran software to generate a mapping table of the target to the camera atmospheric transmittance and solar reflection value under different wavebands, different observation angles and solar azimuth angles, and store the two-dimensional interpolation table in the local txt.

[0026] CUDA parallel calculation of the observation angle and the solar zenith angle of each facet, the first observation angle refers to the included angle between the line of sight direction of the camera observing the facet and the facet normal direction, and the solar zenith angle refers to the included angle between the direction of the sunlight direction pointing to the facet and the facet normal direction. After calculating the observation angle and the solar zenith angle, the corresponding solar reflection radiation brightness of the previous facet is taken out from the two-dimensional interpolation table . According to the solar reflection radiation brightness of the previous facet , the target intrinsic radiation scattering data of each triangular facet is calculated: ; ; Among them, is the radiation intensity of the target current facet, is the intrinsic radiation brightness of the target current facet, is the solar reflection radiation brightness of the target current facet, is the cosine value of the observation angle of the current facet and the facet normal. is the total radiance of the current target, i.e. the intrinsic radiance scattering of the target, which is obtained by superimposing the radiance of all the facets.

[0027] Specifically, in step S101, the background intrinsic radiance scattering data is generated in real time according to the space-time spectrum parameters, the ground / cloud / sky base data and the model library through CUDA parallel calculation, see Figure 2 , including: (1) According to the position of the environmental light source, the position and normal information of the background surface, and the position data of the observation point, the second observation angle and the solar zenith angle of the observation line direction are obtained through geometric relationship calculation; (2) According to the camera viewport parameters and the second observation angle, the latitude and longitude matrix projected onto the earth's surface is obtained through CUDA parallel projection calculation; (3) According to the latitude and longitude matrix, the ground temperature, ground emissivity, cloud temperature, cloud emissivity, ground BRDF and cloud BRDF data are obtained by sampling the background model library; (4) According to the solar zenith angle, the current direct solar radiation is obtained by interpolating the solar irradiance table; (5) According to the ground temperature and ground emissivity, the ground intrinsic radiation is obtained by calculating the Planck radiation law; (6) According to the direct solar radiation and the ground BRDF, the ground solar reflection radiation is obtained by calculating the reflection model; (7) According to the cloud temperature and cloud emissivity, the cloud intrinsic radiation is obtained by calculating the Planck radiation law; (8) According to the direct solar radiation and the cloud BRDF data, the cloud solar reflection radiation is obtained by calculating the reflection model; (9) According to the second observation angle, the tangent height corresponding to each line of sight is obtained by CUDA parallel calculation to obtain the tangent height matrix; (10) According to the tangent height matrix, the sky radiation brightness matrix is calculated; (11) According to the ground intrinsic radiation, the ground solar reflection radiation, the cloud intrinsic radiation, the cloud solar reflection radiation and the sky radiation brightness matrix, the background intrinsic radiance scattering data is obtained by calculating the radiation field synthesis.

[0028] Exemplarily, the background elements include the ground, the atmospheric sky, the cloud, the sun and the like, the physical characteristics of the background elements are provided by the background model library, including time, space and spectral dimension characteristics. The background elements provide the basic background radiation distribution in the whole simulation process, and the high-accuracy background characteristics ensure the simulation accuracy in the imaging simulation process. The background radiation characteristics are modeled as follows: (1) Modeling of the scattering characteristics of ground surface radiation: Based on the ground surface radiation scattering model, Modtran is used to generate the solar ground irradiance and bidirectional reflectance distribution function (BRDF) generated by different second observation angles and solar zenith angles in different wave bands. With the known ground feature cloud layer data and different regional ground feature material emissivity data, the ground surface radiation field is simulated: Based on the infrared radiation scattering model, during the scene simulation process, the observation angle and solar zenith angle of the observation line direction can be calculated in real time according to the position of the environmental light source, the background surface information (position, normal line), and the position of the observation point. Combining the observation angle and solar zenith angle of the line-of-sight direction, the current camera viewport projected onto the surface of the earth is calculated using CUDA parallel computing, and the latitude and longitude matrix is obtained. The ground background basic data (ground temperature, ground emissivity, cloud height) of different environments are sampled through the latitude and longitude. After obtaining the observation angle, solar zenith angle, and different ground background basic data, and the BRDF generated based on the ground surface radiation scattering model, CUDA is used to perform parallel radiation calculation on various background elements: 1) Calculate the solar radiation (solar irradiance to the ground): By sampling the solar irradiance interpolation table, the solar radiation at the current time is obtained by interpolation.

[0029] 2) Calculate the ground background radiation scattering: After obtaining the temperature and other basic data, the ground intrinsic radiation is calculated using the infrared radiation model (Planck formula). Based on the solar radiation data (solar irradiance to the ground) and the ground reflectivity at the current time, the ground solar reflection radiation is calculated.

[0030] 3) Calculate the cloud layer radiation scattering: Similar to step 2), the intrinsic radiation of the cloud layer is calculated. By obtaining the cloud height data, the BRDF data in the BRDF (Bidirectional Reflectance Distribution Function) interpolation table of the cloud layer can be sampled, and then the reflection radiation of the cloud layer to the sun is calculated with the solar radiation data (solar direct radiation) at the current time.

[0031] (2) Modeling of the infrared radiation characteristics of the earth's limb: Based on the infrared radiation model of the limb background, a database of limb background infrared radiation brightness in different wave bands, different altitudes, different latitudes, and different seasons is established using the Modtran software. After calculating the observation angle and solar zenith angle of the observation line direction through the position of the environmental light source, the background surface information (position, normal line), and the position of the observation point, the tangent height matrix of the current camera viewport projected onto the limb surface is calculated using CUDA parallel computing through the observation line direction. The current limb background infrared radiation brightness is obtained by interpolating the tangent height, current time, current wave band, current latitude, and season.

[0032] After the above steps, the target and background infrared radiation scattering simulation data can be obtained and output to the next module.

[0033] S102, based on the target intrinsic radiation scattering data and the generated background intrinsic radiation scattering data, calculate the target entrance pupil characterization radiation and the background entrance pupil characterization radiation. Specifically, in step S102, the radiation characterization before the target entrance pupil is expressed as follows: ; Background radiation before the entrance pupil is represented as: ; in, Indicates the scattering of the target's intrinsic radiation; This represents the atmospheric transmittance from the target to the detector; This represents the characteristic radiation before the entrance pupil at the Earth's surface; This indicates the radiation characterizing the cloud layer before it enters the pupil.

[0034] For example, see Figure 3 The calculation of the radiation characteristic of the target / background before the entrance pupil is roughly as follows: (1) Atmospheric transmission transmittance and atmospheric path radiance calculation: The observation angle and solar zenith angle of the camera (detector) line of sight are calculated according to the method of the target / background intrinsic radiation scattering generation module (1). Based on the known target position and detector position, the atmospheric transmittance from the target to the detector, the atmospheric transmittance from the ground surface and clouds to the detector, and the atmospheric path radiance are calculated using Modtran software in the current band range.

[0035] (2) Complete calculation of the target / background entrance pupil characterization radiation generation module: Characterization radiation before target entrance pupil The simplified calculation formula is as follows: ; in, For the target's intrinsic radiation scattering, The atmospheric transmittance from the target to the detector.

[0036] Background pre-entry pupil characterization radiation is divided into surface pre-entry pupil characterization radiation. Characterization radiation before cloud entry pupil The simplified calculation formula is as follows:

[0037] ; ; in, For intrinsic radiation of the Earth's surface, for intrinsic radiation of clouds, For the scattering of solar radiation by the Earth's surface, For the scattering of solar radiation by clouds, Atmospheric path radiation from the Earth's surface to the detector, Atmospheric path radiation from clouds to the detector, Atmospheric transmittance from the Earth's surface to the detector, Atmospheric transmittance from clouds to the detector, For cloud transmittance, For surface reflectivity, is the bidirectional reflectance distribution function of clouds.

[0038] Characterized radiation before the entrance pupil of the ground and characterized radiation before the entrance pupil of clouds Superposition yields the background radiation characterization before the entrance pupil. .

[0039] ; The radiation calculation process in step (2) above uses CUDA for parallel computation, which can optimize computational efficiency. After the calculation, the final target / background entrance pupil characterization radiation is obtained and passed as input to the next module.

[0040] S103 utilizes CUDA parallel computing to optimize the conversion of the target entrance pupil pre-illumination radiation and the background entrance pupil pre-illumination radiation into detector voltage signals, thereby obtaining a grayscale infrared simulation image.

[0041] Specifically, in step S103, the target entrance pupil pre-characterized radiation and background entrance pupil pre-characterized radiation are converted into detector voltage signals using CUDA parallel computing optimization to obtain a grayscale infrared simulation image, including the following steps S1031 to S1032.

[0042] S1031, The target entrance pupil characterization radiation and the background entrance pupil characterization radiation are superimposed and fused to obtain the input radiation; S1032, Construct an imaging detector model, perform photoelectric detector signal-level imaging modeling on the input radiation, and obtain a grayscale infrared simulation image.

[0043] See here. Figure 4 The imaging detector model includes: an optical system module, a detector module, and a signal processing module; The optical system module is used to perform energy transfer and spatial characteristic conversion operations on the input radiation through optical transmittance, spatial modulation and distortion correction mechanisms to obtain image plane modulated radiation brightness. Specifically, through optical transmittance, spatial modulation, and distortion correction mechanisms, energy transfer and spatial characteristic conversion operations are performed on the input radiation at the entrance pupil to obtain the image plane modulated radiation brightness, including: (1) By adding distortion parameters, the line of sight offset is calculated to obtain the corrected second observation angle and geographic coordinates; (2) Based on the corrected second observation angle, the target entrance pupil characterization radiation and the background entrance pupil characterization radiation are re-performed to obtain the updated target entrance pupil characterization radiation and the background entrance pupil characterization radiation. Based on the updated target entrance pupil characterization radiation and the background entrance pupil characterization radiation, the updated input radiation is calculated. (3) Based on the transmittance database of the optical system, perform band attenuation calculation of the updated input radiation to obtain the attenuated radiation; (4) The attenuated radiation is spatially modulated by MTF to obtain the image plane modulated radiance.

[0044] The detector module is used to perform optical band conversion, spatiotemporal filtering, and noise superposition on the image plane modulated radiance to obtain a noisy detector voltage signal; Specifically, the image plane modulated radiance undergoes optical band conversion, spatiotemporal filtering, and noise superposition to obtain a noisy detector voltage signal, including: (1) Based on the physical parameters of the detector, the image plane modulated radiance is converted by photoelectric conversion to obtain the initial voltage signal; (2) Based on the detector pixel size, the spatial transfer function is calculated to obtain the spatial transfer function, and based on the detector carrier lifetime, the time filter function is calculated to obtain the time filter function; (3) Spatial filtering is performed on the initial voltage signal according to the spatial transfer function and the time filtering function to obtain the noisy detector voltage signal.

[0045] The signal processing module is used to superimpose circuit noise and quantize grayscale on the voltage signal of the noisy detector to obtain a grayscale infrared simulation image.

[0046] Specifically, the circuit noise is superimposed and grayscale is quantized on the voltage signal of the noisy detector to obtain a grayscale infrared simulation image, including: (1) Based on the circuit design parameters, generate circuit noise using Gaussian distribution; (2) The circuit noise is superimposed with the voltage signal of the noisy detector to obtain an infrared simulation image.

[0047] For example, in this invention, after acquiring the characterization radiation before the target / background entrance pupil, it is necessary to establish an imaging detector model, and perform photoelectric detector signal level imaging modeling on the simulated target and background infrared radiation through the satellite staring camera detector. Each step of the calculation process uses CUDA for parallel computing to optimize computational efficiency.

[0048] This invention models and processes three components: the optical system, the detector, and the signal processing circuit, to simulate the image after radiation is transferred to the infrared imaging sensor. After obtaining the target / background entrance pupil characterization radiation, the target and background entrance pupil characterization radiations are first superimposed and fused to obtain the input of the detector response model. : ; in, It is the background radiation characterization before the entrance pupil. It is the radiation characterizing the target before it enters the pupil.

[0049] 1) Optical system: The influence of the optical system model on the input radiation mainly includes the optical system energy attenuation, the total modulation transfer function (MTF) of the optical system, and the optical system distortion.

[0050] 1.1) Optical System Energy Attenuation: First, a database of optical system transmittance (txt files) is generated based on the spectrum and the optical system. During simulation, the optical system is configured to use the optical system specified in this database. It is then determined whether the calculated spectral band is within the operating wavelength range. If not, the transmittance needs to be interpolated. Finally, the obtained radiance is multiplied by the corresponding optical system transmittance to obtain the optical system energy attenuation, as shown in the following formula. For the transmittance of the optical system , Characterize the radiation of the input fusion target background before the entrance pupil. The total radiance obtained after passing through the optical system: ; 1.2) Calculation of the Overall Modulation Transfer Function (MTF) of an Optical System: The MTF of an optical system consists of the MTF of the optical system (transmissive or reflective) and the MTF of the aberrations. The cutoff frequency of the optical system can be calculated from the starting and cutting-off wavelengths and the optical aperture. The aberration MTF can be calculated using the spatial angular frequency, the diffraction-limited cutoff frequency of the optical system, and the root mean square of the aberrations. If the optical system is a transmissive system, the MTF is calculated using the spatial angular frequency and the cutoff frequency of the optical system; if the optical system is a reflective system, the MTF is calculated using the obscuration ratio of the reflective system, the spatial angular frequency, and the cutoff frequency of the optical system. Then, the aberrations... With optical system Multiplying these yields the overall modulation transfer function (MTF) of the optical system. The radiance after passing through the optical system is then calculated. Multiplying by the overall modulation transfer function (MTF) yields the MTF-modulated radiance. As shown in the following formula: ; 1.3) Distortion: By adding distortion parameters to the optical system parameters, a certain distortion variable can be added to the original line of sight, resulting in distortion error between the line of sight calculation and the actual line of sight, directly changing the line of sight of the observation, and thus changing the observation angle in (1), thereby changing the latitude and longitude matrix and height matrix of the line of sight projection onto the ground.

[0051] 2) Detector: The detector divides the input into photoelectric response characteristics calculation, spatial transfer characteristics calculation, and spatial-temporal noise characteristics calculation.

[0052] 2.1) Calculation of photoelectric response characteristics: The radiance obtained by superimposing the effects of various optical systems is calculated. Converted into voltage output using the following formula : ; in, , These are atmospheric transmittance and optical system transmittance, respectively. To detect the pixel area, For detector responsivity, This is the voltage gain coefficient. For the F-number of the optical system, This represents the detector noise voltage value. This represents the base voltage value when the detector has no response.

[0053] 2.2) Spatial transfer characteristic model: The detector's filtering characteristics consist of two parts: spatial filtering and temporal filtering.

[0054] For image area is The spatial angle is , A rectangular detector whose response function is a rectangular box function. Fourier transform to spatial frequency , Separate The product of functions is the detector's level. for: ; The detector itself also contains a time-filtered transfer function, whose modulation transfer function for: ; in, Spatial frequency; This corresponds to the frequency at 3dB. ( (Detector carrier lifetime).

[0055] The voltage output obtained through photoelectric response Voltage output after superimposed spatial transfer characteristics for:

[0056] ; 2.3) Spatial-temporal noise characteristics: This system uses NETD (Noise-Equivalent Temperature Difference) to describe and simulate noise. First, the radiation of the preset NETD temperature plus the background temperature is calculated using Planck's formula. The difference between the NETD temperature and the temperature calculated from the background temperature obtained in (1) is then subtracted to obtain the radiation difference after superimposing the NETD temperature. The obtained radiation difference is converted into voltage to obtain the NETD equivalent noise voltage. Using the NETD noise equivalent voltage as the standard deviation of the Gaussian random number and a preset noise voltage gain coefficient as the mean, a Gaussian function is used. A set of random numbers is generated as NETD noise simulation values ​​to simulate space-time domain noise. The simulated noise values ​​are then superimposed onto the output voltage values ​​obtained in the previous step. Up, obtain voltage output .

[0057] ; 3) Signal processing circuit: Statistical methods (Gaussian distribution) are used to model the circuit noise of different circuit designs. The simulated noise values ​​are then superimposed on the voltage values ​​obtained in the previous step. Up, obtain voltage output .

[0058] ; in, These are the mean and standard deviation of a Gaussian distribution; appropriate values ​​can be selected based on the specific processing circuit.

[0059] Before entering the pupil, the target / background radiation passes through three stages: the optical system, the detector, and the signal processing circuit, and the obtained voltage value is... Perform grayscale conversion and output an 8-bit grayscale value.

[0060] The basic framework of this invention for CPU+GPU collaborative work is as follows: Figure 5As shown. The user end completes the basic parameter configuration; the algorithm layer is divided into two parts: CPU and GPU. The CPU end completes the basic state configuration and core logic processing, including parameter configuration and state parameter updates, such as satellite position, velocity, attitude, and pointing; camera state integration time and frame rate; calculation of sun position and zenith angle; and target position, attitude, and radiation characteristics. Then, it transmits key parameters to the device end. The GPU end uses CUDA to complete intensive data processing and returns relevant state parameters. The two work together to ensure the simulation accuracy and computational efficiency of the system. The visualization end completes the visualization of relevant system results, including two-dimensional image results, three-dimensional situation results, system simulation state parameters, simulation evaluation results, and other information.

[0061] The various embodiments described in this specification are presented in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on its differences from other embodiments. All or part of this invention can be used in numerous general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, mobile communication terminals, multiprocessor systems, microprocessor-based systems, programmable electronic devices, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices, etc.

[0062] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the present invention.

Claims

1. A method for generating simulation signals for space-based end-to-end photoelectric imaging based on CUDA, characterized in that, include: The surface of the target 3D model is divided into multiple triangular facets, and the intrinsic radiation scattering data of each triangular facet is calculated. Based on the spatiotemporal spectral parameters, surface / cloud / edge basic data and model library, background intrinsic radiation scattering data is generated in real time through parallel computing using CUDA. The intrinsic radiation scattering data of the target includes: the intrinsic radiance of each facet and the total radiation intensity of the target. Based on the target intrinsic radiation scattering data and the generated background intrinsic radiation scattering data, the target entrance pupil characterization radiation and the background entrance pupil characterization radiation are calculated. The target pre-entry pupil characterization radiation and the background pre-entry pupil characterization radiation are converted into detector voltage signals using CUDA parallel computing optimization to obtain a grayscale infrared simulation image.

2. The method for generating simulation signals for space-based end-to-end photoelectric imaging based on CUDA according to claim 1, characterized in that, The process of dividing the surface of the target 3D model into multiple triangular facets and calculating the target intrinsic radiation scattering data for each triangular facet includes: Determine the observation parameters for each triangular element; wherein the observation parameters include: element area, element normal, and element temperature; Based on the surface element temperature, the intrinsic radiance of each triangular surface element is obtained using the radiance calculation formula. Based on the solar radiation model, a mapping table is obtained for different observation angles and solar azimuth angles at different wavelengths; wherein, the mapping table is a mapping between the atmospheric transmittance of the target to the camera and the solar reflectance value; CUDA uses the angle between the surface element normal and the camera's viewing direction as the first observation angle for each triangular surface element. CUDA parallelism uses the angle between the surface element normal and the direction of the sunlight pointing towards the surface element as the first vertex angle of each triangular surface element. The solar reflection value of each triangular facet is obtained by looking up the first observation angle and the first zenith angle in the mapping table. The intrinsic radiation scattering data of each triangular facet element is calculated based on the solar reflection value.

3. The method for generating simulation signals for space-based end-to-end photoelectric imaging based on CUDA according to claim 1, characterized in that, The process of generating background intrinsic radiative scattering data in real time using CUDA parallel computing, based on spatiotemporal spectral parameters, surface / cloud / edge data, and a model library, includes: Based on the location of the ambient light source, the location and normal information of the background surface, and the location data of the observation point, the second observation angle and the solar zenith angle in the direction of the observation line of sight are calculated through geometric relationships. Based on the camera viewport parameters and the second observation angle, the latitude and longitude matrix projected onto the Earth's surface is obtained through CUDA parallel projection calculation. Based on the latitude and longitude matrix, surface temperature, surface emissivity, cloud temperature, cloud emissivity, surface BRDF, and cloud BRDF data are obtained by sampling the background model library. Based on the solar zenith angle, the current direct solar radiation is obtained by interpolating the solar irradiance table; The intrinsic radiation of the Earth's surface is calculated using Planck's law of radiation based on the surface temperature and surface emissivity. Based on the direct solar radiation and the surface BRDF, the surface solar reflected radiation is calculated using a reflection model. Based on the cloud temperature and the cloud emissivity, the intrinsic radiation of the cloud is calculated using Planck's law of radiation. Based on the direct solar radiation and the cloud BRDF data, the solar reflected radiation from the clouds is calculated using a reflection model. Based on the second observation angle, the tangent heights corresponding to each line of sight are calculated in parallel using CUDA to obtain the tangent height matrix; The adjacent edge radiance matrix is ​​calculated based on the tangent height matrix. Background intrinsic radiation scattering data are obtained by combining radiation fields based on the intrinsic radiation of the Earth's surface, the solar reflected radiation of the Earth's surface, the intrinsic radiation of the clouds, the solar reflected radiation of the clouds, and the radiance matrix of the adjacent edges.

4. The method for generating simulation signals for space-based end-to-end photoelectric imaging based on CUDA according to claim 1, characterized in that, The radiation characterizing the target before its entrance pupil is represented as follows: ； The background pre-entry pupil characterization radiation is represented as: ； in, Indicates the scattering of the target's intrinsic radiation; This represents the atmospheric transmittance from the target to the detector; This represents the characteristic radiation before the entrance pupil at the Earth's surface; This indicates the radiation characterizing the cloud layer before it enters the pupil.

5. The method for generating simulation signals for space-based end-to-end photoelectric imaging based on CUDA according to claim 1, characterized in that, The process of using CUDA parallel computing to optimize the conversion of the target pre-entry pupil characterization radiation and the background pre-entry pupil characterization radiation into detector voltage signals to obtain a grayscale infrared simulation image includes: The target entrance pupil characterization radiation and the background entrance pupil characterization radiation are superimposed and fused to obtain the input radiation; An imaging detector model is constructed, and photodetector signal-level imaging modeling is performed on the input radiation to obtain a grayscale infrared simulation image.

6. The method for generating simulation signals for space-based end-to-end photoelectric imaging based on CUDA according to claim 5, characterized in that, The imaging detector model includes: an optical system module, a detector module, and a signal processing module; The optical system module is used to perform energy transfer and spatial characteristic conversion operations on the input radiation through optical transmittance, spatial modulation and distortion correction mechanisms to obtain image plane modulated radiation brightness. The detector module is used to perform light band conversion, spatiotemporal filtering and noise superposition on the image plane modulated radiance to obtain a noisy detector voltage signal; The signal processing module is used to perform circuit noise superposition and grayscale quantization on the voltage signal of the noisy detector to obtain a grayscale infrared simulation image.

7. The method for generating simulation signals for space-based end-to-end photoelectric imaging based on CUDA according to claim 6, characterized in that, The process of performing energy transfer and spatial characteristic conversion operations on the input radiation through optical transmittance, spatial modulation, and distortion correction mechanisms to obtain image plane modulated radiation brightness includes: By adding distortion parameters and calculating the line-of-sight offset, the corrected second observation angle and geographic coordinates are obtained. Based on the corrected second observation angle, the target entrance pupil characterization radiation and the background entrance pupil characterization radiation are re-performed to obtain updated target entrance pupil characterization radiation and background entrance pupil characterization radiation. Based on the updated target entrance pupil characterization radiation and background entrance pupil characterization radiation, the updated input radiation is calculated. Based on the transmittance database of the optical system, the band attenuation of the updated input radiation is calculated to obtain the attenuated radiation; The attenuated radiation is spatially modulated using MTF to obtain the image plane modulated radiance.

8. The method for generating space-based end-to-end photoelectric imaging simulation signals based on CUDA according to claim 6, characterized in that, The step of performing optical bandgap conversion, spatiotemporal filtering, and noise superposition on the image plane modulated radiance to obtain a noisy detector voltage signal includes: Based on the physical parameters of the detector, the image plane modulated radiance is photoelectrically converted to obtain an initial voltage signal; Based on the detector pixel size, the spatial transfer function is calculated to obtain the spatial transfer function, and based on the detector carrier lifetime, the time filtering function is calculated to obtain the time filtering function. The initial voltage signal is spatially filtered according to the spatial transfer function and the time filtering function to obtain the noisy detector voltage signal.

9. The method for generating simulation signals for space-based end-to-end photoelectric imaging based on CUDA according to claim 6, characterized in that, The step of performing circuit noise superposition and grayscale quantization on the noisy detector voltage signal to obtain a grayscale infrared simulation image includes: Based on the circuit design parameters, Gaussian distribution is used to generate circuit noise. The circuit noise is superimposed with the voltage signal of the noisy detector to obtain a high-precision infrared simulation image.

Images